RE: Units of Measure

[Arofan Gregory <arofan.gregory@commerceone.com>]

Stephanie:

I think there are different types of "human readable". Yes, anyone can
memorize long lists of codes, but there are more accesible ways of encoding
the same information.

In xCBL, we take the approach of trying to render the less-familiar (and
shorter) EDI code lists in a form that is more intuitive (American English
Words) than pure codes. For the more-familiar lists, we simply use existing
code sets.

This is a tough problem, but unless you have been doing EDI for a long time,
you generally find pure codes daunting. Plus, there are over-loaded codes
across the "standard" lists in the EDI world, which raises the issue of what
a particular alpha-numeric code really represents.

As for who the "humans" are, I would propose the following list:

(1) Developers and the technicians who maintain the system
(2) Small- and Medium-sized enterprises' businesspeople, who might not have
the benefit of translation software, and merely look at the business
messages in a browser or other simple presentation format
(3) Operators of the ystems, responsible for error-handling

The first and last classes can be relied on to learn what is needed. The
middle category is one that did not exist with as great frequency in the
traditional EDI world, but that becomes increasingly important. ebXML's
charter tells us not to forget this group.

Within core components, a proposal has been made that all values are coded
numerically, without any semantic reference, but that all the *tags* in the
documents be an implementation in some specific language (American English,
French, Japanese, what have you). This way, we can track equivalencies
across languages, while still preserving the utility of having
human-readable semantics evident in the "tag names" and "codes", and having
this in a localizable way. I think this is a good approach, if only because
memorizing numbers isn't my strong suit!

Cheers,

Arofan Gregory

-----Original Message-----
From: COOPER,STEPHENIE (HP-PaloAlto,ex1)
[mailto:stephenie_cooper@hp.com]
Sent: Monday, July 10, 2000 9:45 AM
To: ebxml-core@lists.oasis-open.org
Subject: RE: Units of Measure


Hi.

Dumb question of the day:  Have we defined the "human" in "human-readable"?
Who is going to read raw XML?  What category of human?  In the world of X12
and EDIFACT, legacy languages for EDI, the end-user rarely reads the raw X12
or EDIFACT.  The technical support person for EDI may have occassion to read
the raw X12 or EDIFACT when troubleshooting.  Both X12 and EDIFACT are based
on "English", although it's not clear if it's American English, British
English, or some other kind of English.  Presumably, one of the
qualifications to become a technical support person is the ability to read
the raw file and interpret it. 

What will happen in the XML world?  If the goal is machine-to-machine
communication, you've got to codify.  William has adequately pointed out the
pitfalls of free-form text.  For machine-to-person, will there not be tools
available that can take an XML containing codes and tags, point to a
dictionary, and do the interpretation for the person in that equation?

Maybe this is another dumb question, but why be insistent that XML be human
readable?  Doesn't that hark back to when XML was new and the invisioned
usage was primarily web publishing?  Although some may not believe it, I'm
human (;-)).  I can read X12 and EDIFACT.  Can we not conclude from this
that X12 and EDIFACT are human-readable?  Can we not conclude that if we use
tags and codes in XML, it will still be human readable?  Again, who is the
"human" in "human-readable?"

Kind regards,
Steph.

_________________________________________
HEWLETT-PACKARD 
SUPPLY CHAIN INFORMATION SOLUTIONS  
SUPPLY CHAIN ELECTRONIC BUSINESS TEAM             
_________________________________________
Stephenie Cooper, M/S 4L-5                        
eBusiness Development Engineer        
1501 Page Mill Rd. Palo Alto, CA 94304 US         
Tel. 1/650.857-6970,  Fax. 1/650.857.5544 
Internet: stephenie_cooper@hp.com
Websites:  www.hp.com, www.hpebiz.com
_________________________________________
Member, Board of Directors
Electronics Industry Data Exchange Assoc.
www.eidx.org
_________________________________________


-----Original Message-----
From: Rachel Foerster [mailto:rachelf@ix.netcom.com]
Sent: Sunday, July 09, 2000 2:12 PM
To: ebxml-core@lists.oasis-open.org
Subject: RE: Units of Measure


William,

Thanks for the clarification. Norman Walsh's statement which you've quoted
would seem to me to argue against abstract tags that would point to either
an X12 or UN/EDIFACT data element. Additionally, it would also seem to argue
against your belief that a code value from one of these two standards be
conveyed rather than some human-readable definition. But then, as you've
pointed out, look at all of the problems with the variations on how a code
definition could be stated.

This doesn't seem to point to a clear and easy path for on-the-fly
interoperability of XML-based documents, does it?

Rachel

>
> Rachel Foerster asked "Actually, if I understand the XML
> Recommendation
> correctly, it's only the tag that must be human readable, correct?"
>
> Dear Rachel:
>
> One of the requirements in section 1.1 Origin and Goals in the XML 1.0
> Specification, at http://www.w3.org/TR/REC-xml, does say "XML
> documents
> should be human-legible and reasonably clear."  Norman Walsh adds "If
> you don't have an XML browser and you've received a hunk of XML from
> somewhere, you ought to be able to look at it in your favorite text
> editor and actually figure out what the content means."  See XML
> Development Goals at http://www.xml.com/pub/98/10/guide1.html.
>
> I would add that it's reasonable to expect the reader to be
> an expert in
> the problem domain if they're really going to read the XML document;
> probably only somebody intimately familiar with EDI implementation
> guidelines, and EDIFACT and X12 terminology, would understand the tags
> and structure in igML, say; see http://www.igml.org/.
>
> Some have advocated nonsense tags (like BZRTVN) whose meaning
> would have
> to be extracted from some repository, in order to avoid
> offending those
> whose mother tongue is not English.   Unfortunately, those tags would
> clearly violate the human-legibility requirement since most
> people can't
> pronounce anything without vowels.  By the way, I'm still waiting for
> some clown to advocate neutral star dates because ebXML's ISO
> 8601 date
> and time requirement uses the Christian calendar.
>
> I couldn't figure out what John Motley meant by "An X12 message would
> simply be another RENDERING with a different DTD."  What's
> X12 got to do
> with a hunk of XML data? And wouldn't you ever need just one DTD? Why
> would a DTD have to be rendered at all?  It's data that might be
> rendered with various style sheets (or programs) to display the same
> data with French, Spanish, English, or German labels and legends.  Is
> John talking about rendering a DTD's element names in a different
> language?  If so, I can't see the point of that since only programmers
> would in all likelihood ever look at XML directly, and they deal with
> English labels and commands all day long now in HTML, C++, and Java.
>
> William J. Kammerer
> FORESIGHT Corp.
> 4950 Blazer Memorial Pkwy.
> Dublin, OH USA 43017-3305
> +1 614 791-1600
>
> Visit FORESIGHT Corp. at http://www.foresightcorp.com/
> "Commerce for a New World"
>
>